Predicting Individualized Outcomes for Deceased Kidney Donor Waitlisted Candidates and Recipients

Kidney transplantation remains the optimal treatment for end-stage kidney disease (ESKD). However, the persistent disparity between the demand and supply of deceased donor (DD) kidneys underscores the need for better tools to assess transplant outcomes and donor kidney quality. The current Kidney Allocation System (KAS) relies on the Kidney Donor Risk Index (KDRI) to quantify DD kidney quality, yet it combines allograft failure and patient death into a single outcome, limiting its accuracy. In this paper we present refined statistical models to predict post-transplantation risk, differentiating between allograft failure and patient death as competing risks. Using comprehensive data from the Organ Procurement and Transplantation Network/Scientific Registry of Transplant Recipient (OPTN/SRTR) for 2000-2017, our models incorporate biological and clinical factors instead of donor race, account for within-center clustering and between-center variation, and capture non-linear relationships between risk factors. Our results reveal distinct risk factors for allograft failure and patient death. These models provide more personalized risk estimates tailored to donor kidney quality and recipient characteristics, aiding shared decision-making on kidney acceptance. Comparisons with the original KDRI demonstrate the superiority of our separate models, with improved predictability and reduced bias. Our approach eliminates the need to conflate allograft failure and patient death, leading to more accurate risk assessment and better-informed decisions regarding kidney offers. In conclusion, our study underscores the importance of distinguishing between allograft failure and patient death in kidney transplant risk assessment. By offering more precise risk estimates, our models enhance the transparency and efficiency of kidney acceptance decisions, ultimately benefiting both transplant providers and candidates. We also provide a web-based tool to facilitate this process, promoting better outcomes in kidney transplantation. Key PointsO_LIImproved statistical models for kidney transplant risk assessment, separating the risks of allograft failure and patient death. C_LIO_LIModels provide more personalized risk estimates, outperforming the existing Kidney Donor Risk Index (KDRI). C_LIO_LIModels enhance transparency and accuracy in evaluating donor kidney quality, aiding both providers and candidates in decision-making. C_LIO_LIResearch improves the efficiency of kidney acceptance processes, leading to more successful transplants. C_LI